Fiber optic communications systems continue to grow dramatically in terms of numbers, capacity, and complexity. One of the factors responsible for this trend is the increasing sophistication of optical amplifiers, particularly erbium doped fiber amplifiers (EDFA""s). Furthermore, the technology driving EDFA""s is expanding the spectral bandwidth in which such an amplifier can amplify input signals so that scores of channels with sub-nanometer spacing can be amplified and transmitted in current systems. Other network optical elements such as DWDM""s, WADM""s, optical cross-connects, etc., which are compatible for use in communications systems employing 80+channel amplifiers are also now available.
As data rates, bandwidths, and system architectures continue to grow with increasing demand, system performance remains the bottom line criteria, which becomes ever more challenging. EDFA""s, for instance, exhibit characteristic gain spectra which dictate the transition of an attenuated input signal into an amplified output signal. In the days of single or few-channel optical systems, signal transmission could be selected in one or more spectral windows corresponding to flatter portions of the gain spectrum of the optical amplifier, however, for an 80 channel erbium doped fiber amplifier the gain spectrum from 1520 nm to 1565 nm is anything but flat. The unmodified gain spectrum for a typical erbium doped aluminosilicate fiber has a strong peak at about 1532 nm. As such, input channels in the spectral window undergoing significantly greater gain can reach power levels where nonlinear optical effects seriously degrade system performance. On the other hand, signal channels experiencing lower gain will typically exhibit reduced signal to noise ratios, also contributing to degraded system performance.
The non-flat gain spectrum of an erbium doped fiber amplifier is somewhat amenable to flattening through the use of gain flattening filters (GFF""s). A shortcoming to this approach is that even a perfect GFF will only yield a flat gain amplifier spectrum at a specific operating point. As the operating conditions change (i.e., the inversion values change as a result of a change in, e.g., input power, spectral hole burning, pump wavelength drift, etc.) the gain shape will change and therefore the GFF will no longer yield a flat gain spectrum.
It is further possible to compensate for the changing operating conditions discussed above by adjusting the gain shape. This can be accomplished; for example, by tuning the pump wavelength, controlling the temperature of the erbium doped fiber, controlling the pump power, adjusting the amplifier gain via a variable optical attenuator, employing a tunable filter, and through other means appreciated by those skilled in the art. Typically, however, amplified systems that use such control techniques as discussed above are directed at flattening the gain spectrum of the amplifier or of each amplifier in a link having a plurality of amplifiers. In practice, even a finely tuned control scheme in combination with a GFF does not yield a completely flat gain spectrum. Both the control scheme and the GFF introduce some gain flatness errors. Gain or gain flatness sensing applied to an individual amplifier is only capable of detecting and correcting gain flatness errors due to that specific amplifier. The inevitable sensor errors due to optical tap and filter fidelity, local spectral hole burning due to a sensing channel, or other causes, result in a gain flatness error from each amplifier. When such random sensing errors cause several amplifiers in a link to tilt their gain spectra in a way that happens to correlate, large overall link power flatness and optical signal to noise ratio degradations appear. If gain flatness of each amplifier is optimized, these degradations can only be avoided by using extremely precise gain flatness sensors. Even if the amplifiers are identical and each is individually optimized for flatness, all gain flatness errors will accumulate. Thus power flatness errors due to control errors accumulate with each amplifier in the link. Accordingly, the inventors realized a need for improving optical amplifier, network optical element, and system link performance beyond that obtained by gain spectrum flattening for maintaining and improving overall optical communication system performance.
This objective and others, and the advantages associated therewith can be obtained according to the present invention in which the output power spectrum of an amplifier, network optical element, or an optical network link using these components, is preferentially shaped, and in particular, flattened to provide a flattened input power spectrum to an immediately following amplifier or component or to a receiving apparatus.
In accordance with a first embodiment of the invention, in an amplified, multiwavelength, multichannel optical transmission system which includes a network transmission link having N (N greater than 1) optical amplifiers each of which supplies gain to an attenuated input signal spectrum for outputting an amplified signal spectrum, a method of tuning and improving the performance of the system includes the steps of detecting a characteristic of an optical signal (e.g., a peak power and a noise value) at the end of the link (e.g., at the output of the Nth amplifier) and controlling the gain spectra of amplifiers in the link so as to maximize the optical signal to noise ratio (OSNR) at the end of the link.
In another embodiment of the invention, a method for tuning and improving the performance of an optical link in an optical transmission system which includes one or more optical amplifiers involves monitoring the output power spectrum of each amplifier And modifying the gain spectra of the amplifier as it operates on an input power spectrum to the amplifier to provide a flattened output power spectrum. In an aspect of this embodiment, the output power spectrum is flattened by optimizing the lowest OSNR for all of the output channels of the amplifier.
Another embodiment of the invention includes an amplifier, or a network optical element providing at least some equalization between input and output signals, or a network link comprising the optical amplifier and/or network optical element in which the optical amplifier and/or the network optical element and/or the end of the link has an output power spectrum profile that is flattened, and flatter than the profile of the respective input power spectrum. In an aspect of this embodiment, an exemplary amplifier is an EDFA, and exemplary network optical elements include DWDM""s, WADM""s, optical cross-connects, and others. These exemplary devices are by no means intended to limit the invention as any active or passive device capable of producing a modified output power spectrum is suitable.
In an aspect of all of the foregoing embodiments, a portion of the output from the respective device or link is diverted via a coupler or the like into a device such as an optical spectrum analyzer, which is coupled to a means for modifying the gain spectrum or an equalization spectrum of the respective device. Exemplary embodiments of techniques for modifying the gain or equalization spectra include gain tilt, input signal power level, input signal spectral profile, pump wavelength tuning, and/or generally modifying the inversion level of an active fiber.
In all of the foregoing embodiments, the per channel output power of each amplifier is preferably maximized subject to a maximum channel power (PChannel) and total output power (xcexa3PChannel), which are determined empirically to avoid penalties due to optical nonlinearities in the transmission fiber.